This application outlines a program of research that continues our work on psychosocial influences on health, emphasizing the importance of studying multiple psychosocial risk factors in the context of socioeconomic status (SES) and other demographic variables. Four studies are proposed that address several questions. All will target significant clinical events such as coronary events and mortality as the primary dependent variables. Two of these are designed to test aspects of a model of mechanisms that account for the connections between stress and health outcomes. One of these is the notion that tangible, social, and psychological resources can moderate stress reactions. This leads to the hypothesis that good social relationships can be especially protective for those of low SES. This will be tested in a large representative population sample (Study One) and a large sample of military veterans. A corollary of this hypothesis is that the interaction of social support and SES should be most apparent in those under high levels of job strain, which will also be tested in the population sample. Both physical health and depressive symptoms will be the outcomes. Another resource is effective coping strategies for stress producing conditions. One of these may be tension reduction through moderate alcohol consumption, a proposition will be examined as an interaction between moderate alcohol usage and job strain in the same sample. Alcohol consumption is also hypothesized to affect health through other pathways in the conceptual model. We have found that some psychosocial variables (hostility and depression) are associated with patterns of binge drinking and alcohol-related behavior problems, although this effect may be moderated by SES. This could account for some of the associations between hostility, depression, and health, a proposition that will be tested in the sample of veterans. The other studies will concentrate on the task of better illuminating the health effects due to particular components of two complex psychosocial variables that are often treated as more global constructs. One study of cardiac patients will rely on an extensive battery of depressive symptom measures collected during hospitalization to predict subsequent cardiac events and mortality. The other is a study of community volunteers who were assessed with a comprehensive set of hostility measures in the early 1990's. These will serve as predictors of subsequent cardiac events and mortality. All studies are expansions of our past and current work, building on those findings to test new hypotheses, extend notions based previous findings to the prediction of significant clinical outcomes, and improve our conceptualizations of widely studied psychosocial risk factors. [unreadable] [unreadable] [unreadable]